Deep-Sea Research II 50 (2003) 1085–1105

Upper-ocean currents around Taiwan

W.-D. Lianga, T.Y. Tanga,*, Y.J. Yangb, M.T. Koa, W.-S. Chuanga

a Institute of Oceanography, National Taiwan University, Taipei, Taiwan, Republic of Chinab Department of Marine Science, Chinese Naval Academy, Kaohsiung, Taiwan, Republic of China

Accepted 12 October 2002

Abstract

Current velocity, measured by Shipboard Acoustic Doppler Current Profiler (Sb-ADCP) during 1991–2000, was used

to study the upper-ocean (o300 m) currents around Taiwan. The collected data were debugged, calibrated, grid, and

averaged to compose a three-dimensional current-velocity distribution. The validity of the composite current velocity

was supported by 12 sets of moored current-velocity time series. Qualitative agreement was obtained. The moored time

series also indicated that the seasonal variation of current around Taiwan was generally weak except for the shallow-

water regimes.

The composite and moored currents revealed a branch of the Kuroshio that intruded steadily and persistently into

the South China Sea. Part of the intruded Kuroshio flowed out of the South China Sea through the northern Luzon

Strait and re-united with the main stream Kuroshio. The Kuroshio had two velocity maximum cores southeast of

Taiwan, but gradually combined into one as the Kuroshio flowed north. The Kuroshio was deflected by the I-Lan

Ridge east of Taiwan and the zonal-running shelf break northeast of Taiwan. At the shelf break, the Kuroshio split,

with one branch intruding onto the shelf.

West of the Luzon Strait, the Kuroshio intruded into the South China Sea. Some water flowed northward into the

Taiwan Strait and re-joined the Kuroshio. Currents in the Taiwan Strait flowed primarily in a northward direction,

except for the southward current near the coast of Mainland China. North of the Taiwan Strait, a branch of the

northward flow followed the northern coast of Taiwan to join the Kuroshio.

The composite current varied consistently from season to season. There was generally poor correlation between

currents and local winds, especially in the deep-water regime. Remote forces were important in the currents around

Taiwan.

r 2003 Elsevier Science Ltd. All rights reserved.

1. Introduction

Taiwan is an island located in the tropical tosub-tropical western Pacific (Fig. 1). East ofTaiwan, where the shelf is narrow and the wateris deep, lie Green Island, Yonaguni Island and the

islands of Lan-Yu. The I-Lan Ridge connectsYonaguni Island to Taiwan. Northeast of Taiwan,the shelf break extends northeastward, and turnsabruptly toward the zonal direction around 26�N.The North Mien-Hwa Canyon, which stretchesacross the shelf break around the turning point,connects the shelf of the East China Sea with thedeep ocean basin. The shelf north of Taiwan isbroad. The Taiwan Strait between Taiwan and

*Corresponding author.

E-mail address: tyt@ccms.ntu.edu.tw (T.Y. Tang).

0967-0645/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved.

doi:10.1016/S0967-0645(03)00011-0

Mainland China measures 200 km in width,400 km in length, and 60 m in average depth. Onlyat the southeastern portion of the Taiwan Strait, isthe water deeper than 200 m. The funnel-shapedPeng-Hu Channel is located between Taiwan andthe Peng-Hu Archipelagos. West of Peng-HuChannel is the shallow Taiwan Bank, whichconnects the northern shelf of the South ChinaSea. The Luzon Strait between Taiwan and Luzonis a passage between the western Pacific Ocean andthe South China Sea. Its width is around 400 km

and its maximum depth is 2000 m. On both sides ofthe Luzon Strait, the water is deep.

The predominant wind system around Taiwan isthe monsoon. The northeasterly monsoon prevailsNovember–March, and the southwesterly mon-soons prevail from May–September. April andOctober are transition months. (Chuang andLiang, 1994). The monsoon system is the primaryforce causing current variation around Taiwan(Wyrtki, 1961). Nitani (1972) described surfacecurrents around Taiwan in summer and winter (see

118˚E

118˚E

120˚E

120˚E

122˚E

122˚E

124˚E

124˚E

18˚N 18˚N

20˚N 20˚N

22˚N 22˚N

24˚N 24˚N

26˚N 26˚N

-8000

-7000

-6000

-5000

-4000

-3000

-2000

-1000

0m

East China Sea

South China Sea

Mainland China

Taiwan

Luzon

Taiwan

Stra

it

Taiwan Bank

Luzon Strait

NorthMien-Hwa

Canyou

Wu-C

hiu Is

.

Yonaguni Is.

I-Lan RidgeP

eng-

Hu

Cha

nnel

Pen

g-H

uA

rchi

pela

gos

Green Is.

Lan-Yu Is.

M1

M2

M3

M4

M5

M6M7

M8M9

M10M11

M12

Fig. 1. Bathymetric chart showing areas around Taiwan and mooring locations.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–11051086

Fig. 2). The Kuroshio flows to the north along thecoast of Taiwan and the shelf break of the EastChina Sea. In winter, the Kuroshio intrudes intothe South China Sea in the Luzon Strait. Insummer, water from the South China Sea entersthe western edge of the Kuroshio. East of Taiwan,the width of the Kuroshio is around 100 km, witha maximum speed of around 100 cm s�1. TheKuroshio’s pathway parallels the shoreline ofTaiwan and demonstrates no obvious seasonalvariation. South of Peng-Hu Archipelagos, currentin the Taiwan Strait flows predominantly north-eastward regardless of the seasons. North of thePeng-Hu Archipelagos, the current flows north-eastward in summer and southwestward in winter.The seasonal variation is significant only in theLuzon Strait and in the northern Taiwan Strait.

Many oceanographers believe that the Kuroshiointrudes into the South China Sea through theLuzon Strait in winter and stops in summer. Thiswas first described by Wyrtki (1961), and has been

re-confirmed by a number of studies (e.g., Shaw,1989). There are, however, a few studies (e.g., Quet al., 2000; and Li and Wu, 1989), that indicatethat the intrusion of the Kuroshio also occurs insummer. The potential intrusion of the Kuroshiothrough the Luzon Strait is controversial. Priorstudies were largely based upon hydrographicmeasurements. This study will provide directcurrent velocity measurements and clarify the issueof Kuroshio intrusion.

Based upon hydrographic measurements andgeostrophy, the width, maximum speed, andtransport of the Kuroshio east of Taiwan havebeen estimated in a number of earlier studies. Thefindings were similar, but contained some notabledifferences. For example, the transports estimatedby Nitani (1972), Chu (1976), and Liu (1983)varied from 19 to 47 Sv. The estimated depth ofthe Kuroshio varied from a few hundred to a1000 m. Sun (1987) indicated that the main axis ofthe Kuroshio varied seasonally, especially north of

Fig. 2. Current charts in knots: (a) summer, and (b) winter (after Nitani, 1972).

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–1105 1087

the I-Lan Ridge, moving close to the coast inwinter and away from the coast in summer. Chu(1976) found that the Kuroshio had no clearannual signal but did have a large interannualvariation. Recently, moored current velocity mea-surements along the I-Lan Ridge and altimetryobtained from the TOPEX/POSEDIENT satelliteindicated that westward propagating eddies fromthe Pacific Ocean had a significant impact on theKuroshio (Yang et al., 1999; Zhang et al., 2001).The transport of Kuroshio reveals significantintra-seasonal variation, with a time scale of 100days, but with annual variation that is vague.

Analysis of the moored current velocity, tem-perature, and salinity time series northeast ofTaiwan, Tang et al. (2000) suggested that the axisof the Kuroshio migrates not only seasonally, butalso intra-seasonally. Such migration has a greatimpact on the flow pattern north of Taiwan, wherethe Kuroshio intrudes onto the shelf and interactswith the Taiwan Strait outflow. A southwestwardcountercurrent, flowing opposite of the Kuroshio,was observed along the shelf edge northeast ofTaiwan (Chuang and Wu, 1991; Tang and Yang,1993; Chuang et al., 1993). As the Kuroshio movesclose to shore, it intrudes directly onto the shelfalong northeastern Taiwan. The northwestwardcurrent prevails in the upper ocean (o150 m) andthe countercurrent is submerged. When the Kur-oshio moves away from the shore, the Kuroshiointrudes onto the shelf primarily through theNorth Mien-Hwa Canyon. A cyclonic eddy formsbetween the North Mien-Hwa Canyon and Tai-wan (Tang et al., 1999). The countercurrentsurfaces to complete the annual cycle. Using theShipboard Acoustic Doppler Current Profiler (Sb-ADCP), Tang et al. (2000) pointed out that thecurrent on the shelf north of Taiwan is primarilynorthward flowing, regardless of season. Theimpact of monsoon on the current is minimal.

Chu (1961) claimed that the current in TaiwanStrait is influenced by a number of factors,including the intrusion of the Kuroshio, the SouthChina Sea current, monsoon, and the Chinacoastal current. Wang and Chern (1988) foundthat the current in the Taiwan Strait was swift andnorthward under the southwesterly monsoon insummer, while the Mainland China coastal current

flowed southward in the northern Taiwan Straitunder the northeasterly monsoon. The northwardwarm and salty Kuroshio water meet the cold,fresh China coastal water around the centralTaiwan Strait and a thermohaline front forms,largely prohibiting the Kuroshio from flowingfarther north (Wang and Chern, 1989). Chuang(1986) found that the current in the Peng-HuChannel was northward almost year around. Themonsoon does not reverse the current directionexcept under persistent strong northeasterly mon-soon.

This study explores the distribution of three-dimensional upper (o300 m) ocean current velo-city around Taiwan. The current velocity datacollected during 1991–2000 from the R/Vs OceanResearcher I, II, and III with the Sb-ADCP wereanalyzed. Twelve sets of moored current velocitydata were used to verify the composite currentvelocity and to infer the temporal variation of thecurrent. This paper will proceed as follows. Section2 describes the data and the composite currentvelocity distribution around Taiwan. Section 3discusses the qualitative comparison between thecomposite current velocity and the moored currentvelocity measurements. Section 4 investigates thethree-dimensional upper (o300 m) ocean currentvelocity around Taiwan in vertical-meridianal orvertical-zonal sections of current velocity. Adiscussion and summary are provided in Section 5.

2. Data and composite current velocity

The Sb-ADCP measurements used in this studyhave been provided by the Ocean Data Bank/National Center for Ocean Research (ODB/NCOR) of National Science Council (NSC),Republic of China (ROC). Three Taiwaneseresearch vessels, Ocean Researcher I, II, and IIIcollected the Sb-ADCP measurements from 1991to 2000. In general, the Sb-ADCP was set tomeasure the current velocity every second and thento record the ensemble average every 1–2 min. Thedepth bin was 8–16 m, and the depth of range ofmeasurement was from 16 to 320 m. The globalpositioning system (GPS) recorded the ship loca-tion. The ship velocity was estimated based on the

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–11051088

position time series. To obtain the compositecurrent velocity, the data were processed asfollows. The current velocity was first debugged,calibrated, and aligned using the method describedby Tang and Ma (1995). To reduce the randomerror of estimated ship velocity, which wascalculated from the ship positions recorded bythe GPS, a 30-min running average was applied.The current velocity also was averaged every30 min. Hereafter, each averaged current velocityis treated as a single data point. The mean shiptrack during that 30 min was used as the locationof averaged current velocity. The root meansquare (rms) error associated with the calibratedand averaged current velocities was less than73.5 cm s�1. Finally, the velocities were linearlyinterpolated vertically with 10 m intervals andaveraged horizontally within the grid of size of0.25� � 0.25�. No further interpolation or smooth-ing was applied.

Since the Sb-ADCP measurements were notevenly distributed, the number of data pointsvaried in each grid varied. The averaged current

velocity in each grid is more representative whenthere is more data. Conversely, the more datarequired within each grid, the more fragmented theobtained composite current velocity distributionbecomes. A choice has to be made between theminimum requirement of data and integration ofcomposite current velocity. The semidiurnal tidalcurrent is the primary high-frequency fluctuationto bias the composite current velocity in each grid.The semidiurnal tidal velocity has an amplitude of50 cm s�1. Similar to the Sb-ADCP data, the30 min-averaged tidal velocity, %u; was calculated.To randomly choose N pieces of %u; its average, %uN ;would be the noise causing the difference betweenthe composite current velocity and true meancurrent velocity. Theoretically, %uN should decreaseas N increases. Finally, to repeat the experiment100 times, the rms of %uN is calculated. Possibleerrors caused by the semidiurnal tidal current inthe composite current velocity have been esti-mated. Fig. 3 shows the distribution of numbers ofthe data, the plot of rms value of %uN versus the N;and the available number of grids versus N: The

1~5

6~10

11~50

51~100

101~500

>500

118E 120E 122E 124E18N

20N

22N

24N

26N

0 %

20 %

40 %

60 %

80 %

100 %

0 5 10 15 20 25 30N

Ava

ilab

le g

rid

0

5

10

15

20

25

30

35

40

0 5 10 15 20 25 30

rms

of u

N (

cm/s

)

Fig. 3. The left panel denotes the distribution of numbers of data. The right-upper and right-lower panels represent the rms value of

the %uN and the available percentage of grid versus the N; respectively.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–1105 1089

numbers of the data in the grids were distributedunevenly. In general, the grid further away fromthe Taiwan shoreline had less data. The grids hadmaximum data in the regions of research vesselharbors, at the northern tip and southwestern tipof Taiwan. The rms value deceases rapidly with N

increase initially, but gradually flattens out. Theavailable number of grids also decreases with N

increase. To remain the relatively integrateddistribution of composite current velocity, N ¼ 6was chosen as the minimum requirement. Theestimated bias caused by the semidiurnal tidalcurrent would be 14 cm s�1, around 1

4of tidal

amplitude. Therefore, we limit our description anddiscussion on the composite current velocitydistribution to qualitative comments.

Fig. 4 shows the composite velocity vectors at 30and 100 m around Taiwan. In general, the currentsat these tow-depths showed no significant differ-

ence. North of Luzon, the Kuroshio flowed to thenorthwest, and turned clockwise, forming a clock-wise path in the Luzon Strait. The main path ofthe Kuroshio returned to the east of Taiwan, butsome Kuroshio water intruded into the northernSouth China Sea. West of Luzon, the currentflowed northeastward and turned counterclock-wise to the west as it encountered the Kuroshio.The intruding Kuroshio water and the SouthChina Sea water moved west (seen more clearlyat 100 m). The westward current seemed to splitinto two branches. The southern branch wasdiverse and vague, perhaps due to insufficientdata. Results from numerical models (e.g., Shawand Chao, 1994; Metzger and Hurlburt, 2001)indicate that this is part of the South China Seacirculation, which flows along the edge of the shelfinto the central basin of South China Sea. Thenorthern branch (Fig. 4) turned clockwise and

118˚E

118˚E

120˚E

120˚E

122˚E

122˚E

124˚E

124˚E

18˚N

20˚N

22˚N

24˚N

26˚N

100 cm/sec

At 30 m Depths

118˚E

118˚E

120˚E

120˚E

122˚E

122˚E

124˚E

124˚E

18˚N

20˚N

22˚N

24˚N

26˚N

-8000

-7000

-6000

-5000

-4000

-3000

-2000

-1000

0m

At 100 m Depths

Fig. 4. Composite of the Sb-ADCP current velocity vectors around Taiwan. The left and right panels represent the 30 and 100 m

depths, respectively.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–11051090

toward the Kuroshio along the southern tip ofTaiwan or else flowed into the Taiwan Straitthrough the Peng-Hu Channel, producing a clock-wise circulation pattern south of the Taiwan Strait.

The current in the Kuroshio is swift along theEast Coast of Taiwan. Its maximum speed andwidth were around 100 cm s�1 and 100 km, respec-tively. Over the I-Lan Ridge, the Kuroshio turnednortheastward and left the coast of Taiwan. Only aweak flow was deflected eastward by the I-LanRidge, indicating that the I-Lan Ridge did notblock the Kuroshio. The Kuroshio flowed north-eastwardly along the shelf break after it leftTaiwan. Around 26�N, the zonal shelf breakblocked the Kuroshio, and the Kuroshio turnedeastward. Some water entered the shelf. On theshelf north of Taiwan, current was weak in theregion around 25.3–25.8�N, 121.7–122.2�E. Out-flow of the Taiwan Strait was observed along thenorthern coast of Taiwan. North of this region, thecurrent velocity was disorderly and confused. Thisclutter distribution might be caused by theinsufficient data and the large bias induced bythe tidal currents on the shelf. Tang et al. (2000)claimed that the region of weak current wasvisible. Because of the seasonal migration of theKuroshio, the upper (o150 m) ocean currentsreversed during the seasons of northeasterly andsouthwesterly monsoon, resulting in a weakcomposite current.

The currents converged at the funnel-shapedPeng-Hu Channel and flowed northward in theTaiwan Strait. As the northward current passedover the northern opening of the Peng-HuChannel, a ridge separated it into two branches.One branch was deflected westward, following thetopography. The other branch flowed over theridge to the north. Currents with double peaks inthe Taiwan Strait were observed north of the ridge,but became vague farther north. Some outflowfrom the Taiwan Strait followed the northerncoast of Taiwan into the Kuroshio. The majorityof the Taiwan Strait outflow moved to the north,interacting partially with the intruding Kuroshio.

The composite current velocities at 30 m undersouthwesterly and northeasterly monsoons areshown in Fig. 5. During southwesterly monsoon,the flow pattern was similar to the previous

velocities distribution. The lack of data in thenortheasterly monsoon season was primarily theresult of rough sea conditions prohibiting researchvessel activities. Current-velocity distributions inthe two seasons were similar except a few notabledifferences. The Kuroshio entered the Luzon Straitprimarily during the southwesterly monsoon. Onlya small amount of water entered the South ChinaSea. The Kuroshio intrusion in the Luzon Straithad a larger incident angle during the season ofnortheasterly monsoon than during the season ofsouthwesterly monsoon. Northeast of Taiwan, asmall cyclonic flow seen on the edge of the shelf(centered around 25.5�N, 122.5�E) was observedduring the season of southwesterly monsoon. Thecyclonic flow disappeared and the Kuroshiomoved close to the shelf when the northeasterlymonsoon prevailed. In the Taiwan Strait, thecurrent velocity was weaker during the north-easterly monsoon than during the southwesterlymonsoon, but the flow was always northeastward.No southward current was observed.

3. Comparison of moored and shipboard current-

velocity measurements

Twelve sets of moored current-velocity timeseries, measured by the self-contained AcousticDoppler Current Profiler (ADCP), are used toaffirm the validity of composite current velocity.The seasonal and intra-seasonal variations incurrent velocity are noted.

Fig. 6 shows four sets of daily current velocitystick time series at M1, M2, M3, and M4,locations of which are shown in Fig. 1. M1–M3are in the central portion of Luzon Strait and theM4 is around 230 km west of M2. The currentvelocity was most often at 100 m. The secondsegment at M2 was at 140 m because of a missedcount in the mooring rope length. The duration ofthe measurement at each site was longer than 5months. In general, the current velocities at varieddepths correlated well in the upper 200 m of thewater column, but their amplitude decreased withthe depth. The current velocity at M1 and M2 wereessentially northwestward, but M1 was morenorthward than M2. The transition of monsoons

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–1105 1091

from southwest to northeast usually occurred inOctober (Chuang and Liang, 1994), and had littleimpact on the current velocity. The current at M1and M2 indicated that the Kuroshio flowed intothe South China Sea steadily and persistently.Differing from the current at M1 and M2, thecurrents at M3 were weak and varied, frequentlyfrom northeastward to northwestward. The cur-rents alternated in and out of the South China Sea.The southern boundary of Kuroshio intrusioncould be around M3. The current velocity at M4also was weak and varied, but it only changeddirection from northwest to southwest. A west-ward component current velocity was consistentlyobserved. The Kuroshio might intrude further tothe west.

The variability and mean of current velocitiesobtained from the moored and shipboard ADCPat each location were compared (Fig. 7). One

diagram contains the 30-min average Sb-ADCPcurrent velocity in the grid where the mooringoccurred. The other diagram contains the hourlymoored current velocity. The best agreement wasobtained at M1, where the current primarilyflowed northwest or north-northwest. The meanmoored current velocity vector and compositecurrent velocity are nearly identical. At M2, themoored current velocity primarily flowed north-west while the Sb-ADCP current velocity wasmore diverse. However, both current velocitydirections varied in the same range, from northto west (second quadrant). The mean and compo-site current velocity vectors had similar direction,but the former had larger amplitude. Similarly, thetwo types of measurement at M3 had some similarcharacteristics but also had notable differences.Both of them showed they were weaker and hadmore fluctuations than the current at M1 and M2.

118˚E

118˚E

120˚E

120˚E

122˚E

122˚E

124˚E

124˚E

18˚N

20˚N

22˚N

24˚N

26˚N

100 cm/sec

Southwesterly Monsoon

118˚E

118˚E

120˚E

120˚E

122˚E

122˚E

124˚E

124˚E

18˚N

20˚N

22˚N

24˚N

26˚N

-8000

-7000

-6000

-5000

-4000

-3000

-2000

-1000

0m

Northeasterly Monsoon

Fig. 5. Composite of the Sb-ADCP current velocity vectors at 30 m depth around Taiwan. The left and right panels represent the

seasons of southwesterly monsoon and northeasterly monsoon, respectively.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–11051092

Considering the difference in the two samplingmethods, it is reasonable to conclude that the twotypes of measurement agreed qualitatively.

Fig. 8 shows the three daily current-velocitystick time series. The one located at the I-LanRidge had the shortest record (less than 2 months).The other two are located at the shelf breaknortheast of Taiwan. The velocities were at 30 m.At M5, the current was quite stable and flowed tothe north-northeast. A similar observation withmuch longer time series also showed that thecurrent velocity was generally stable, except whenthe westward-propagated eddy collided with theKuroshio. The collision occurred about every 100days (Yang et al., 1999; Zhang et al., 2001). At

M6, the current was first southwestward, and thenchanged to become northwestward by mid-Octo-ber. The migration of Kuroshio caused thevariation of current velocity. Tang and Yang(1993) and Chuang et al. (1993) described themoored current velocities at M6 in detail. At M7,the current velocity was mainly northeastward, butgenerally became more eastward in October–April.Tang et al. (2000) claimed the Kuroshio impingedonto the East China Sea shelf around M7. Themigration of the Kuroshio also caused the varia-tion of current at M7. However, such variationwas not large. In the above three moored currentvelocities, the large seasonal variation was onlyseen at M6.

Fig. 6. The stick diagrams of the 36-h low-pass filtered current velocity. The upper to lower panels represent the time series data of M1

in 1997, 1998, and 1999, M2 in 1997 and 1998, M3 in 1997, and M4 in 2000. The current velocity data was primarily at 100 m depths

except the M2 in 1998, which was at 140 m depth.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–1105 1093

The comparison between the moored and ship-board ADCP current velocity is shown in Fig. 9.In spite of the record being shorter than 2 months,the moored current velocity at M5 agreed wellwith the shipboard ADCP current velocity in bothvariability and mean. At M6, the two types ofmeasurement showed a relatively large disparity;though they also had some similarities. Theirvariability was roughly in the same range. The

mean and composite current velocity vectors hadsmall amplitude and similar direction. The mooredand shipboard ADCP current velocity at M7agreed well in both variability and mean. Gen-erally speaking, the agreement between the twotypes of measurement was good when the currentvelocity variation was small, as occurred at M5.When the current velocity had large fluctuations,as at M6, the disparity between the two types of

0 20 40 60 80 100 >100

Speed (cm/s)

15%

30%N

E

S

W

NE

SESW

NW

15%

30%N

E

S

W

NE

SESW

NW

15%

30%N

E

S

W

NE

SESW

NW

15%

30%N

E

S

W

NE

SESW

NW

Composite

15%

30%N

E

S

W

NE

SESW

NW

15%

30%N

E

S

W

NE

SESW

NW

15%

30%N

E

S

W

NE

SESW

NW

15%

30%N

E

S

W

NE

SESW

NW

Moored

M4

M3

M2

M1

100 cm/sCompositeMoored

Fig. 7. The rose diagrams and mean current velocities obtained from shipboard and moored ADCP measurement at M1, M2, M3, and

M4. The left, central, and right panels represent the rose diagrams of the shipboard and moored ADCP data, and the mean velocities,

respectively. The range of rose is 0–30% with 7.5% interval.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–11051094

measurement increased. But overall, the shipboardADCP current velocity properly represents thecurrent velocity—at least, qualitatively.

Fig. 10 shows the five sets, M8, M9, M10, M11,and M12, daily moored current-velocity time seriesin Taiwan Strait. The velocities were at 30 m. Thefirst four time series were obtained from a mooringarray deployed across the Taiwan Strait. Thecurrent was observed in the duration of mid-fall toearly winter when the northeasterly monsoonprevailed. M12 was located at the southern open-ing of the Peng-Hu Channel. The current wasobserved from June—when the southeasterlymonsoon prevailed to November, when the north-easterly monsoon was dominant. In general, thecurrent velocities at M8, M9, M10, and M11fluctuated considerably in the several day timescale. The coherence between the local wind, whichwas northeasterly, and current velocity wascalculated and generally poor (not shown). AtM11, the current flowed even against wind formost of the observed period. However, the impactof local wind on the current velocity was still seen

by peak-to-peak comparison. The current velocityat M12 primarily flowed with wind moving towardthe southwest. The coherence between the currentvelocities at various stations was also poor. Theabove result implies that the local wind was not theonly dominant external force in the Taiwan Strait;for example, the intruded Kuroshio also couldplay an important role. At M12, the currentvelocity persistently flowed into the Peng-HuChannel, whether the local wind was southwes-terly or northeasterly. The northeasterly monsoon,which intensified after October, only causedfluctuation in the current velocity. Again, the localwind poorly correlated with the current velocity.The former inference was reconfirmed: bothremote and local forces were important in theTaiwan Strait.

The comparison between the moored and Sb-ADCP current velocities in the Taiwan Strait isshown in Fig. 11. Strong agreement between thetwo types of measurement was found at M11 andM12. At M8, M9, and M10, a large disparity wasobserved. The disparity could arise from a number

Fig. 8. The stick diagrams of the 36-h low-pass filtered current velocity. The upper to lower panels represent the time series data of M5

in 1992, M6 in 1991, and M7 in 1992, 1993, and 1994. The current velocity data was at a depth of 30 m.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–1105 1095

of factors. For example, the data were sparse,especially in the western Taiwan Strait. The Sb-ADCP data were lacking when the northeasterlymonsoon was large. There was no moored currentvelocity during the southwesterly monsoon season.The large semidiurnal tidal current velocity alsocould be a source of error for the compositecurrent velocity.

The composite current velocity might inade-quately represent the current where there was alarge current-velocity variation. However, thistype of current is limited around Taiwan. Theabove comparison between moored and Sb-ADCPcurrent velocity indicates that the compositecurrent velocity obtained from the Sb-ADCPmeasurements represents the current around Tai-wan qualitatively, but not quantitatively.

4. Vertical sections of current velocity

Fig. 12 shows the vertical sections of zonal (U)and meridional (V ) components of current velocityalong the Luzon Strait and along 21�N. Thelatitude-vertical section of U along the LuzonStrait showed that the eastward/westward currentalternately occurred from northern Luzon tosouthern Taiwan. The westward current occupieda larger area and had larger amplitude than theeastward current. The largest westward currentoccurred in the central-northern Luzon Strait,while the largest eastward current appeared at thesouthern tip of Taiwan. The northward currentdominated in the V section. Its maximum speedwas over 50 cm s�1. The southward current wasonly observed at the tips of northern Luzon and

0 20 40 60 80 100 >100

Speed (cm/s)

30%

60%N

E

S

W

NE

SESW

NW

30%

60%N

E

S

W

NE

SESW

NW

30%

60%N

E

S

W

NE

SESW

NW

Composite

30%

60%N

E

S

W

NE

SESW

NW

30%

60%N

E

S

W

NE

SESW

NW

30%

60%N

E

S

W

NE

SESW

NW

Moored

M7

M6

M5

100 cm/sCompositeMoored

Fig. 9. Same as Fig. 7 except at M5, M6, and M7. The range of rose is 0–60% with 15% interval.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–11051096

southern Taiwan. The westward current, indicat-ing that the water intruded into the South ChinaSea from Pacific Ocean, generally accompanied thelarge northward current. The intruded currentshould primarily be the Kuroshio. The SouthChina Sea outflow at southern Luzon Strait couldretroflex back to the South China Sea though thecentral Luzon Strait, but its amount was small.The zonal volume transport, calculated by simplyintegrating the U along the section, was around�3 Sv to the west. A smaller, but also negative,value was obtained by using the Sb-ADCPmeasurement collected only in the season ofsouthwest monsoon. Apparently, the Kuroshiointruded into the South China Sea even in thesouthwest monsoon season. Although the trans-port of Kuroshio intrusion was only estimatedqualitatively, the moored current velocity sup-ported such an inference because the Kuroshiointrusion occurred persistently. The earlier beliefthat no Kuroshio intrusion occurred in the LuzonStrait in summer might need to be reconsidered.

Along 21�N, the longitude-vertical section of U

was primarily westward of the Luzon Strait

signifying that the intruded Kuroshio flowedcontinuously westward. The maximum westwardspeed was around 30 cm s�1. East of the LuzonStrait, the U was fragmented, with eastward/westward current reciprocally shown. The north-ward current dominated in the V section. A weaksouthward current was observed west of 119�E.East of this southward current, the northwardcurrent extended from 119�E to 121.75�E. Theregion was wide—around 300 km. The maximumnorthward current, around 60 cm s�1 was in theLuzon Strait. East of this northward current, theV was negative/positive reciprocally. The Kur-oshio became vague around the Luzon Strait,possibly because the western boundary current isdeficient in its western continental boundary.Local topography also might play a role. A fewislands, which lie near the 122�E parallel with theLuzon Strait, might cause the spatial currentfluctuation east of Luzon Strait.

East of Taiwan, four longitude-vertical current-velocity sections along 22�N, 23�N, 24�N, and25�N are shown in Fig. 13. The Kuroshio domi-nated all four sections. On the southernmost

Fig. 10. The stick diagrams of the 36-h low-pass filtered current velocity. The upper to lower panels represent the time series data of

M8, M9, M10, and M11 in 1999, and M12 in 1996. The current velocity data was at a depth of 30 m.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–1105 1097

section (along 22�N), the U was weak. It wasprimarily positive (eastward) between Taiwan andLan-Yu near 121.5�E. East of Lan-Yu, the U

gradually turned to negative (westward). A weakeastward component was seen again east of

124�N. The V was chiefly positive (northward)and had two maximum velocity cores located ontwo sides of Lan-Yu, respectively. The westerncore with maximum speed over 90 cm s�1 was nearthe surface. The eastern core with maximum speed

0 20 40 60 80 100 >100

Speed (cm/s)

30%

60%N

E

S

W

NE

SESW

NW

30%

60%N

E

S

W

NE

SESW

NW

30%

60%N

E

S

W

NE

SESW

NW

30%

60%N

E

S

W

NE

SESW

NW

30%

60%N

E

S

W

NE

SESW

NW

Composite

30%

60%N

E

S

W

NE

SESW

NW

30%

60%N

E

S

W

NE

SESW

NW

30%

60%N

E

S

W

NE

SESW

NW

30%

60%N

E

S

W

NE

SESW

NW

30%

60%N

E

S

W

NE

SESW

NW

Moored

M12

M11

M10

M9

M8

100 cm/sCompositeMoored

Fig. 11. Same as Fig. 9 except at M8, M9, M10, M11, and M12.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–11051098

around 70 cm s�1 was at subsurface (70 m). Theeastward current generally accompanied the wes-tern core while the westward current accompaniedthe eastern core. This implies that the two coreseventually combined together as the Kuroshioflowed further north. The feature of doublemaximum velocity cores could be caused by thelocal topography or the fact that the splitKuroshio in the Luzon Strait had not fully re-united. The southwestward countercurrent wasobserved close to Taiwan at subsurface as well aseast of Kuroshio.

The velocity section along 23�N was similar to italong 22�N. The Kuroshio had double velocitycores. The western core had greater speed and wasat surface, while the eastern core had smaller speedand was at subsurface. Again, the cores seemed tomerge further north. The subsurface counter-current close to Taiwan was barely observed. Westof the Kuroshio, the countercurrent was weakerthan that at the section of 22�N.

The two velocity cores merged into a single coreat the section of 24�N, south of I-Lan Ridge. TheU and V were mostly positive, but the V hadmuch greater amplitude. The negative U and V

were only seen at subsurface close to Taiwan. TheKuroshio was mainly northward; its eastwardcomponent was small. I-Lan Ridge is quite closeto the section of 24�N, but its impact on theKuroshio was not observed here.

The velocity at section of 25�N, north of I-LanRidge, showed that the U and V were stillprimarily positive, but the amplitude of U

increased and the amplitude of V decreased. Thissuggests that the Kuroshio was deflected as itflowed over the I-Lan Ridge. The countercurrentwas observed close to Taiwan. Its south-ward component velocity extended to thesurface, indicating the Kuroshio departed fromTaiwan coast. Such separation could be dueto the deflection of Kuroshio. Using thesimilar data, Tang et al. (2000) found that the

VU

Along 120.75˚E Along 21˚N

0 m

100 m

200 m

300 m

19˚N 20˚N 21˚N 22˚N

0 m

100 m

200 m

300 m

19˚N 20˚N 21˚N 22˚N

0 m

100 m

200 m

300 m

119˚E 120˚E 121˚E 122˚E 123˚E 124˚E

0 m

100 m

200 m

300 m

119˚E 120˚E 121˚E 122˚E 123˚E 124˚E

Fig. 12. Vertical sections of composite Sb-ADCP current velocity of zonal (U) and meridional (V ) components along the 120.75�E

and 21�N are show in left and right panels, respectively. The depth range is from the surface to about 300 m. The continental shelf is

darkened. The values are contoured at intervals of 10 cm s�1. The zero contour appears in heavy line. Portions of positive values are

shaded.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–1105 1099

Kuroshio migrated seasonally, but the migrationwas small.

Fig. 14 shows the width, main axis, and upper-ocean (o300 m) northward volume transport ofKuroshio. The 10 cm s�1 isotach of northwardvelocity was used as an indicator for the bound-

aries of the Kuroshio. The maximum northwardspeed defined the main axis of the Kuroshio. Thelinear integration was applied to estimate thenorthward volume transport. From south tonorth, the main axis of the Kuroshio was closeto the coast of Taiwan south of I-Lan Ridge. It

Alo

ng 2

2˚N

Alo

ng 2

3˚N

Alo

ng 2

4˚N

Alo

ng 2

5˚N

U V

0 m

100 m

200 m

300 m

121˚E 122˚E 123˚E 124˚E

0 m

100 m

200 m

300 m

0 m

100 m

200 m

300 m

0 m

100 m

200 m

300 m

121˚E 122˚E 123˚E 124˚E

0 m

100 m

200 m

300 m

121˚E 122˚E 123˚E 124˚E

0 m

100 m

200 m

300 m

0 m

100 m

200 m

300 m

0 m

100 m

200 m

300 m

121˚E 122˚E 123˚E 124˚E

Fig. 13. Vertical sections of composite Sb-ADCP current velocity of zonal (U) and meridional (V ) components along the 22�N, 23�N,

24�N, 25�N east of Taiwan are show in left and right panels, respectively. The depth range is from the surface to about 300 m. The

continental shelf is darkened. The values are contoured at intervals of 10 cm s�1. The zero contour appears in heavy line. Portions of

positive values are shaded.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–11051100

departed from the coast as it flowed over the I-LanRidge. The maximum speed (>100 cm s�1) alongthe main axis of the Kuroshio showed littlevariation. The width of the Kuroshio was around170 km, but narrowed to 120 km on I-Lan Ridge.The northward volume transport of Kuroshiodemonstrated no significant and systematic changesouth of I-Lan Ridge. It reduced as the Kuroshioapproached and flowed over the I-Lan Ridge. Thereduction of the northward transport of Kuroshiowas essentially caused by deflection of Kuroshio.A part of northward transport turned to eastwardtransport. The values of northward volume trans-port are listed as a reference only since thecomposite current velocity might be inappropriateto make a quantitative estimate. However, thepresent estimated volume transport in the I-LanRidge agreed well with the transport estimated byJohns et al. (2001), whose estimate was obtainedfrom an array of moored current velocity timeseries.

In summary, the Kuroshio flowed primarilynorthward along the eastern coast of Taiwan. Eastof southern Taiwan, the Kuroshio split, but

gradually re-united. As the Kuroshio flowed overthe I-Lan Ridge, it was deflected eastwardly. Thevelocity became more eastward. The Kuroshiodeparted from the Taiwan coast. A subsurfacecountercurrent close to the island was consistentlyobserved. It surfaced north of I-Lan Ridge.

Since the previous studies indicated that thecurrent along the shelf break northeast of Taiwanhad noticeable seasonal variations, Fig. 15 showsthe U and V sections along 25.5�N in the seasonsof southwesterly and northeasterly monsoon. Theseasonal variation was observed. Regardless of theseasons, a negative U occurred around shelf break.It clearly separated the main stream of Kuroshioand the current on the shelf. Corresponding to thisnegative U ; the V was positive and negative in theupper and lower water column, respectively.However, this distribution of V varied seasonally.The negative V occurred in the deep water in thenortheasterly monsoon season, but it extendedupwardly to near surface in the southwesterlymonsoon season. This result agrees with the earlierfinding (Tang et al., 2000) that the subsurfacesouthwestward countercurrent at the shelf break

100 cm/sec

20.8 Sv

20.6 Sv

24.3 Sv

20.0 Sv

18.4 Sv

16.5 Sv

15.4 Sv

119˚E

119˚E

120˚E

120˚E

121˚E

121˚E

122˚E

122˚E

123˚E

123˚E

124˚E

124˚E

125˚E

125˚E

21˚N 21˚N

22˚N 22˚N

23˚N 23˚N

24N˚ 24˚N

25˚N 25˚N

26˚N 26˚N

Fig. 14. The upper ocean (o300 m) northward volume transport of Kuroshio and composite Sb-ADCP current velocity vectors at

30 m depth east of Taiwan. Bold and dashed lines indicate the width and main axis of Kuroshio.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–1105 1101

northeast of Taiwan was submerged when thenortheasterly monsoon intensified. East of thecountercurrents, both U and V were positive. Thenortheastward Kuroshio was presented andshowed much more complicated spatial velocitydistribution than it did east of Taiwan. Thelocations of maximum velocity cores of U and V

were different, implying that the Kuroshio wasbranching. The influence of the zonal-runningshelf break north of this section on the Kuroshiowas noted. A part of the Kuroshio in the upperocean flowed northward or even northwestwardintruding onto the shelf. The shelf blocked thesubsurface Kuroshio, which flowed northeastwardor even eastward. The Kuroshio also variedsignificantly with monsoons. In the southwesterlymonsoon season, the Kuroshio was weak and itsmaximum speed core was at subsurface. In thenortheasterly monsoon, the weak Kuroshio in-tensified and the subsurface core nearly disap-peared. However, the feature of the Kuroshiohaving greater speed in winter than summer couldbe meaningless if we give consideration to the

seasonal migration. When the Kuroshio moved offshore in the southwesterly monsoon season, it maybe that only a portion of the Kuroshio wasobserved. Such migration also caused moreKuroshio intrusion northwestwardly onto theshelf in the northeasterly monsoon season.

On the shelf north of Taiwan, the current wasgenerally complicated. The complicated flow dis-tribution could be related to the intrusion ofKuroshio, Taiwan Strait outflow, and their inter-action. However, the large semidiurnal tidalcurrent velocity on the shelf could result in a largeerror in the computed composite current velocity.Such an error could further increase the complica-tion of current velocity distribution. Therefore,there is no detailed description of the spatialcurrent velocity distribution on the shelf. Never-theless, the current on the shelf seems to preferflowing northeastward regardless of season. Thelocal wind could not be the only dominant forceon the shelf north of Taiwan.

Fig. 16 shows four velocity vertical sectionsalong 22�N, 23�N, 24�N, and 25�N in the Taiwan

VU

Southwesterly Monsoon Northeasterly Monsoon

0 m

100 m

200 m

300 m

119˚E 120˚E 121˚E 122˚E 123˚E 124˚E

0 m

100 m

200 m

300 m

119˚E 120˚E 121˚E 122˚E 123˚E 124˚E

0 m

100 m

200 m

300 m

119˚E 120˚E 121˚E 122˚E 123˚E 124˚E

0 m

100 m

200 m

300 m

119˚E 120˚E 121˚E 122˚E 123˚E 124˚E

Fig. 15. Vertical sections of composite Sb-ADCP current velocity of zonal (U) and meridional (V ) components along the 25.5�N in

seasons of southwesterly monsoon and northeasterly monsoon are show in left and right panels, respectively. The depth range is from

the surface to about 300 m. The continental shelf is darkened. The values are contoured at intervals of 10 cm s�1. The zero contour

appears in heavy line. Portions of positive values are shaded.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–11051102

Strait. On the southern opening of Taiwan Strait(section along 22�N), the V was positive andnegative in the western and eastern portions,respectively, while the positive U dominatednearly the entire section. This type of velocitydistribution supports the former finding, a clock-wise circulation south of Taiwan Strait. In thePeng-Hu Channel, the current velocity was mainlynorthward in the deep channel where the U wasweak. West of the deep channel, the U was

positive and had relatively simple distribution.The V varied from positive to negative and then topositive again. A couple of maximum speed coreswere near bottom. The Peng-Hu Archipelagosnorth of this section could be responsible for thiscomplicated flow distribution. However, the netnorthward transport was positive, indicating thatthe water flowed into the Taiwan Strait. Along thesection of 24�N, the U was negative at the shallowend of the funnel-shaped Peng-Hu Channel, whichcurved northwestward. It was positive at two sidesof the end. The V was primarily positive, except inthe region close to Mainland China. The region’slarge V and negative U nearly coincided, indicat-ing that the Peng-Hu Channel conducted the waterflow into the Taiwan Strait. Near the northernopening of the Taiwan Strait (section of 25�N),both U and V were principally positive. Theirnegative values were seen in the region close toMainland China. Two regions demonstratedrelatively large velocity. One was close to theTaiwan coast. The other one was near the bottomeast of Wu-Chiu Island. West of Wu-Chiu Island,a weak southwestward current was observed.

5. Discussion and summary

Generally speaking, the Kuroshio around Tai-wan demonstrated moderate temporal variation,but complicated spatial distribution, especiallyaround the Luzon Strait and northeast of Taiwan.The spatial complication was caused primarily bylocal topography. Ignoring variations caused bywestward propagating eddies, the observed currentvelocity at I-Lan Ridge (Johns et al., 2001) did notvary with time significantly. The Kuroshio showedsmall temporal fluctuations also confirmed by themoored current velocity in Luzon Strait. Thisresult implies that the Kuroshio around Taiwanmight be stable in time. In contrast to the velocity,the Kuroshio migrated intra-seasonally and sea-sonally. The seasonal migration was especiallyobvious north of I-Lan Ridge. Such seasonalmigration caused large temporal variations in themoored current velocity at shelf break northeast ofTaiwan.

Alo

ng 2

2˚N

Alo

ng 2

3˚N

Alo

ng 2

4˚N

Alo

ng 2

5˚N

U V

0 m

100 m

200 m

300 m

119˚E 120˚E 121˚E

0 m

100 m

200 m

300 m

0 m

100 m

200 m

300 m

0 m

100 m

200 m

300 m

119˚E 120˚E 121˚E

0 m

100 m

200 m

300 m

119˚E 120˚E 121˚E

0 m

100 m

200 m

300 m

0 m

100 m

200 m

300 m

0 m

100 m

200 m

300 m

119˚E 120˚E 121˚E

Fig. 16. Same as Fig. 13 except in Taiwan Strait.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–1105 1103

Because of insufficient data, cross-section ofseasonal current distributions was not presented.Rough seas frequently prohibited the activities ofresearch vessels in winter. The Sb-ADCP datawere less in the northeasterly than the south-westerly monsoon seasons. Inevitably, the compo-site current velocity was biased favoring thecurrent-velocity distribution during the southwes-terly monsoon season. Nevertheless, the mooredcurrent velocity showed that the current aroundTaiwan had little seasonal variation, especially inthe deep-water region. In the shallow regions, suchas Taiwan Strait, the composite current-velocityvectors showed those similar tendencies as themean moored current-velocity vectors obtainedduring mainly the northeasterly monsoon. There-fore, the bias could not change the distribution ofcomposite current significantly.

The qualitative, but not quantitative, validity ofcomposite current was supported by twelve sets ofmoored currents. The results showed that theKuroshio intruded into the South China Seathrough the northern-central Luzon Strait regard-less of season, but its incident angle could varywith season. The Kuroshio with double velocitycores was seen east of southern Taiwan, but thetwo cores gradually combined into one before theI-Lan Ridge, which deflected the Kuroshio. TheKuroshio branched as it collided with the zonal-running shelf break northeast of Taiwan. Themain stream of the Kuroshio flowed along theshelf break to the east and a part of the Kuroshiointruded into the shelf. A submerged counter-current between the Kuroshio and the shelf breakwas consistently observed. The seasonal migrationof the Kuroshio influenced the countercurrent andthe current northeast of Taiwan.

The intruding Kuroshio water in the SouthChina Sea interacted with the South China Seacurrent. A clockwise flow pattern was seen south-west of Taiwan. Most of the water in thisclockwise current returned to the Kuroshio at thetip of southern Taiwan. The rest of the waterflowed into the Taiwan Strait through the Peng-Hu Channel and is an important resource forcurrent in the Taiwan Strait. The current in theTaiwan Strait primarily flowed to the north. Acountercurrent was faintly visible along the coast

of Mainland China. The Taiwan Strait outflowinteracted with the intruding Kuroshio north ofTaiwan. Only a small portion of outflow enteredthe Kuroshio along the northern coast of Taiwan.The rest of outflow continued north.

Acknowledgements

The research was supported by NSC of ROCunder grants NSC 89-2611-M-002-031-OP2 for T.Y. Tang, NSC 90-2611-M-012-001-OP2 for Y. J.Yang, and NSC 89-2611-M-002-023-OP2 for W.-S. Chuang. We express our thanks to the ODB/NCOR for the historical Sb-ADCP and bathyme-try data. The assistance of the captains and crewsof the R/Vs ORI, II, and III are also greatlyappreciated. The plotting software of curvedvelocity vectors was kindly provided by Dr. D. S.Ko at Naval Research Laboratory, Stennis SpaceCenter (NRLSSC).

References

Chu, T.-Y., 1961. On ocean currents in East China Sea and the

Variability of Taiwan. Meteorological Bulletin 4, 7–17 (in

Chinese with English abstract).

Chu, T.-Y., 1976. Study of the kuroshio current between the

Taiwan and Ishigakijima. Acta Oceanographica Taiwanica

13, 140–153.

Chuang, W.-S., 1986. A note on the driving mechanisms of

current in the Taiwan Strait. Journal of Oceanography

Society Japan 42, 355–361.

Chuang, W.-S., Liang, W.-D., 1994. Seasonal variability of

intrusion of the Kuroshio water across the continental shelf

northeast of Taiwan. Journal of Oceanography 50, 531–542.

Chuang, W.-S., Wu, C.-K., 1991. Slope-current fluctuation

northeast of Taiwan, winter 1990. Journal of Oceanography

Society Japan 47, 185–193.

Chuang, W.-S., Li, H.-W., Tang, T.Y., Wu, C.-K., 1993.

Observation of the countercurrent on the inshore side of the

Kuroshio northeast of Taiwan. Journal of Oceanography

49, 581–592.

Johns, W.E., Lee, T.N., Zhang, D., Zantopp, R., Liu, C.-T.,

Yang, Y., 2001. The Kuroshio east of Taiwan: moored

transport observations from the WOCE PCM-1 array.

Journal of Physical Oceanography 31, 1031–1053.

Li, L., Wu, B., 1989. A Kuroshio loop in South China Sea? On

circulation of the northeaster South China Sea. Journal of

Oceanography in Taiwan Strait 8, 89–95.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–11051104

Liu, C.-T., 1983. As the Kuroshio turns: (I) Characteristics

of the current. Acta Oceanographica Taiwanica 14,

88–95.

Metzger, E.J., Hurlburt, H.E., 2001. The nondeterministic

nature of Kuroshio penetration and eddy shedding in the

South China Sea. Journal of Physical Oceanography 31,

1712–1732.

Nitani, H., 1972. Beginning of the Kuroshio. In: Stommel, H.,

Yoshida, K. (Eds.), Kuroshio, its physical aspects. Uni-

versity of Tokyo Press, Tokyo, pp. 129–163.

Qu, T., Mitsudera, H., Yamagata, T., 2000. Instruction of the

North Pacific waters into the South China Sea. Journal

Geophysical Research 105, 6415–6424.

Shaw, P.-T., 1989. The instruction of water masses into the sea

southwest of Taiwan. Journal Geophysical Research 94,

18213–18226.

Shaw, P.-T., Chao, S.-Y., 1994. Surface circulation in the South

China Sea. Deep-Sea Research 41, 1663–1683.

Sun, X., 1987. Analysis of the surface path of the Kuroshio in

the East China Sea. In: Sun, X. (Ed.), Essays of the

Investigation of Kuroshio. Ocean Press, Beijing, pp. 1–14

(in Chinese with English Abstract).

Tang, T.Y., Yang, Y.J., 1993. Low frequency current variability

on the shelf break northeast of Taiwan. Journal of

Oceanography 49, 193–210.

Tang, T.Y., Ma, J.C., 1995. A note on the accuracy of

shipboard ADCP on Ocean Researcher I. Acta Oceano-

graphica Taiwanica 34, 71–81.

Tang, T.Y., Hsueh, Y., Yang, Y.J., Ma, J.C., 1999. Continental

slope flow northeast of Taiwan. Journal of Physical

Oceanography 29, 1353–1362.

Tang, T.Y., Tai, J.H., Yang, Y.J., 2000. The flow pattern north

of Taiwan and migration of the Kuroshio. Continental Shelf

Research 20, 349–371.

Wang, J., Chern, C.-S., 1988. On the Kuroshio branch in the

Taiwan Strait during wintertime. Progress in Oceanography

21, 469–491.

Wang, J., Chern, C.-S., 1989. On cold water instructions in the

eastern Taiwan Strait during the cold season. Acta

Oceanographica Taiwanica 21, 469–491.

Wyrtki, K., 1961. Physical oceanography of the southeast Asia

waters. Scripps Institute of Oceanography NAGA Report,

Vol. 2, 195pp.

Yang, Y., Liu, C.-T., Hu, J.-H., Koga, M., 1999. Taiwan

Current (Kuroshio) and impinging eddies. Journal of

Oceanography 55, 609–617.

Zhang, D., Lee, T.N., Johns, W.E., Liu, C.-T., Zantopp, R.,

2001. The Kuroshio east of Taiwan: Modes of variability

and relationship to interior ocean mesoscale eddies. Journal

of Physical Oceanography 31, 1054–1074.

W.-D. Liang et al. / Deep-Sea Research II 50 (2003) 1085–1105 1105